INDEX

Explanations

phrases related to document revisions or updates

oai_token-act-pair · gpt-3.5-turbo

instances of the word "to" indicating changes or updates

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 dearly

-0.66

Reviewer

-0.65

icious

-0.65

boxing

-0.61

 exited

-0.61

icide

-0.60

awar

-0.59

zin

-0.59

icho

-0.58

 hatched

-0.58

POSITIVE LOGITS

 accommodate

1.59

 reflect

1.50

 coincide

1.28

 resemble

1.21

 include

1.20

 incorporate

1.19

 conform

1.19

 allow

1.14

 exclude

1.13

 maximize

1.13

Activations Density 0.149%

phrases related to document revisions or updates

instances of the word "to" indicating changes or updates

No Comments

No Known Activations